Administrative Volatility and Immigration Enforcement: The Structural Sources of Instability in a Static Statute

I. Introduction

Throughout decades of political change, the text of the Immigration and Nationality Act has remained largely settled in its core enforcement architecture, while the experience of immigration law has not. Since the substantial amendments of the Illegal Immigration Reform and Immigrant Responsibility Act of 1996 and the REAL ID Act of 2005, the statute’s grounds of removal and its basic allocation of authority have stood relatively still [1]. A noncitizen whose removal case was pending in 2017 confronted one enforcement regime. The same case, still pending in 2021, confronted another, and by 2025 it may well have confronted a third. Nothing in the governing statute compelled these shifts. Instead, they arrived through executive instruments, some taking the form of presidential orders and agency memoranda, and others taking the form of Attorney General precedent decisions or of guidance that is formally nonbinding but operationally decisive. The law changed, in short, while its text stood still.

This evaluation therefore asks a question that the existing scholarship has raised but not answered in an integrated way: why does enforcement policy shift so sharply across presidential administrations when the governing statute does not change, and what does that instability do to the fairness of adjudication? Two subsidiary questions follow from it. First, what legal and institutional features make enforcement policy inexpensive to reverse? Second, through what mechanisms does that reversibility travel from a signed memorandum to an individual case? The premise from which the inquiry proceeds is that the condition just described is no accident, and that volatility in immigration enforcement is a function of legal structure. To begin, the structure does its work through delegation, in that the Act confers broad authority over enforcement implementation while providing almost no mechanism that requires continuity across administrations. Adding to this, it does its work through dependence, since the immigration courts remain housed within the Department of Justice, an arrangement that places adjudication under the supervision of politically appointed leadership. Furthering this, it does its work through insulation, judicial doctrine having largely shielded enforcement prioritization from sustained review, so that the principal levers of change are difficult to contest through litigation. Collectively, these features create an environment in which enforcement policy can be reorganized rapidly and with limited procedural friction. Stability, where it exists, is a function of political continuity, and owes little to any structural commitment that would secure it.

The effects of this architecture are visible in the system’s outputs. Most prominently, the immigration court backlog has moved at historic scale in both directions, expanding to levels without precedent across the period from fiscal year 2017 through fiscal year 2024 and then contracting during fiscal year 2025 in the first sustained decline in more than a decade [2], [3]. Asylum grant rates have moved comparably, falling from a peak above fifty percent in September 2023 to under twenty percent by August 2025, all during a period in which the statutory standard for asylum remained unchanged [4], [5]. Part IV.A sets out these figures in full, along with the source and measurement cautions that attach to them. What matters at the outset is the direction of the inference they support. These figures reflect more than administrative congestion, illustrating instead that enforcement standards and adjudicatory outcomes respond to executive recalibration, and that individuals whose cases span transitions encounter a system in which the posture of the administration at the moment of decision does real work alongside statutory eligibility and factual merit.

Two definitions frame the inquiry, and they are worth fixing at the outset because the argument returns to them repeatedly. Stability, as used here, means continuity in the operative standards that shape enforcement and adjudication over time, and it does not mean uniformity of results. Fairness means predictability and the consistent treatment of similarly situated individuals. Because the second definition does substantial work in Part V, it requires a defense rather than a stipulation, and the rule-of-law tradition supplies one. Lon Fuller, a legal philosopher who taught at Harvard Law School, identified congruence between declared rules and their administration, along with constancy of law through time, as internal conditions without which a system fails to qualify as law at all rather than merely as bad law [29]. Joseph Raz, a legal philosopher associated with Oxford and Columbia, adds the complementary point that the rule of law is valuable precisely because it enables people to plan, and that its violation frustrates reliance regardless of whether the substantive rules are just [30]. Collectively, these accounts illustrate that predictability is not a preference smuggled into the definition of fairness, but an independently defended criterion by which legal systems are assessed. That grounding matters because it locates the problem in the structure of governance and treats the particular enforcement philosophy of any given administration as beside the point. A restrictionist administration and an expansionist one can each, in principle, run a stable system. Neither can run a fair one if the ground rules shift beneath cases already in motion.

To ground the inquiry, this evaluation combines quantitative analysis with original qualitative evidence drawn from practitioner fieldwork. The quantitative component identifies where outcomes move and when, while the qualitative component explains how those movements are produced on the ground, drawing on a key informant interview with an experienced immigration attorney, identified throughout as [Redacted] and described only in deliberately broad terms to preserve confidentiality. The terms of that protection, including the informant’s review of quoted material, are set out in Part III. As the informant put it, “policies can change overnight.” The observation describes a system in which enforcement governance is conducted through instruments that require neither notice-and-comment procedures nor a transition period, illustrating how statutory delegation and institutional informality combine to produce instability that is experienced less as deliberate policymaking than as legal unpredictability.

Five substantive parts follow. Part II reviews the relevant scholarship and case law and isolates the gap this evaluation addresses, namely the absence of an integrated account connecting the design choices that make reversal easy to the measurable volatility that reversal produces. Part III explains the explanatory sequential mixed-methods design and the handling of the practitioner evidence. Part IV reports the results of both phases, presenting the quantitative indicators and the practitioner-reported mechanisms as findings. Part V discusses those findings, tracing them to the doctrinal and institutional architecture that produces them, assessing what they mean for fairness, and developing four reforms designed to raise the cost of reversal without denying the Executive lawful space to prioritize enforcement. Part VI concludes.

II. Literature Review

Immigration law has long been understood as a statutory domain in which Congress supplies the basic architecture while the Executive supplies much of the operative content through charging choices, docket management, and rulemaking. That arrangement is not novel. What is newly salient is the degree to which modern immigration governance appears to turn on administrative reversals that follow leadership transitions, often with immediate downstream consequences for people whose cases remain pending. A large literature has noticed this pattern. Fewer accounts have isolated the legal and institutional mechanisms that make rapid reversibility routine, and fewer still have linked those mechanisms to measurable volatility in adjudicatory outcomes over time. This Part therefore reviews the relevant work in stages, beginning with the doctrinal baseline and the limited friction that administrative law supplies, then turning to the scholarship on presidential administration, to the institutional literature on adjudication, and to the empirical record, before closing with background allocation rules and a statement of the gap this evaluation fills.

A. Discretion and Its Insulation: The Doctrinal Baseline

To begin, the doctrinal starting point predates immigration law’s modern controversies. In Heckler v. Chaney, the Supreme Court held that an agency’s decision not to enforce is presumptively unreviewable, reasoning that enforcement choices involve a balancing of priorities and resources that courts are poorly positioned to second-guess [6]. That presumption supplies the background rule against which immigration enforcement operates, illustrating that the decision to act, or not to act, against a given individual sits at the far edge of what courts will examine.

Adding to this, immigration doctrine builds directly on that baseline. In Reno v. American-Arab Anti-Discrimination Committee, the Court framed the commencement of removal as a paradigmatic exercise of prosecutorial discretion and construed 8 U.S.C. § 1252(g) to bar review of decisions to commence proceedings, adjudicate cases, or execute removal orders [7], [8]. The significance of the decision extends beyond the mere existence of discretion. Rather, discretion is treated as structurally embedded in the statutory scheme and insulated from ordinary review, which supplies an enabling condition for enforcement variability across administrations without statutory revision. Connected to this, the Court’s federal primacy cases reinforce the same architecture from another angle. In Arizona v. United States, the Court emphasized that removal reflects a nationally coordinated system requiring priority-setting under scarcity, and it invalidated most of Arizona’s effort to construct a parallel enforcement regime on that basis [9]. By conceptualizing enforcement allocation as a core feature of federal governance, the reasoning entrenches the Executive as the system’s practical manager.

Continually, a more recent line of cases addresses the reviewability of enforcement guidelines themselves. In United States v. Texas, the Court held that the plaintiff states lacked standing to challenge federal removal priorities, and several opinions stressed the institutional limits of judicial supervision over priority-setting [10]. Whatever one’s view of the result, the case illustrates the practical consequence of justiciability doctrine for stability, in that if enforcement guidance is difficult to challenge in court, then rapid shifts in guidance can occur with limited judicial friction. Collectively, Heckler, Reno, Arizona, and Texas function less as discrete holdings than as a standing permission structure for enforcement volatility.

B. The Friction That Does Exist, and Why It Falls Short

Opposingly, reversal is not entirely frictionless, and any honest account of the literature must say so. In Department of Homeland Security v. Regents of the University of California, the Court set aside the rescission of the Deferred Action for Childhood Arrivals program because the agency failed to supply a reasoned explanation or to consider reliance interests [11]. Similarly, in Biden v. Texas, the Court permitted an administration to unwind its predecessor’s border program, holding that the governing statute did not compel the program’s continuation while treating the memorandum effecting the change as final agency action subject to the Administrative Procedure Act [12]. Collectively, these cases illustrate that administrative law does impose procedural discipline on reversal.

Crucially, however, the discipline is of a limited kind, and the scholarship has begun to recognize as much. Review of this sort polices the paperwork of reversal rather than the fact of it, so that an administration which explains itself adequately may still unwind its predecessor’s regime in full. Adding to this, the review is episodic and slow, arriving years after the challenged guidance has already reshaped thousands of pending cases, and it rarely reaches the memoranda and internal directives through which day-to-day recalibration occurs. Where review does reach further, as when the courts of appeals reject Attorney General precedent, it proceeds case by case and long after the fact. Ultimately, procedural review slows the pendulum without shortening its arc. This distinction between friction on the manner of change and friction on the frequency of change is underdeveloped in the existing literature, and it matters for any stability-oriented reform agenda.

C. Presidential Administration and Enforcement as Policymaking

Furthering this, doctrinal discretion becomes policy formation in the hands of Presidents. In a pair of leading articles, Adam Cox and Cristina Rodríguez, professors of law at New York University and Yale respectively, argue that the modern President is the principal architect of immigration policy, using enforcement tools such as deferred action and priority guidance to generate policy effects that are functionally durable even when Congress cannot legislate [13], [14]. Their core contribution is to make presidential policy influence appear predictable rather than exceptional, and to be specific about instruments, illustrating that the statutory structure itself makes enforcement tools unusually central to what immigration law means in practice. Connected to this, Hiroshi Motomura, a professor of law at the University of California Los Angeles, reaches a compatible conclusion from a different direction, showing that the gap between the statute’s formal breadth and its practical enforcement leaves millions of people living under a regime defined more by executive tolerance than by legal status [15]. Similarly, Shoba Sivaprasad Wadhia, a professor of law at Pennsylvania State University, completes the picture from both directions, as her account of prosecutorial discretion traces its long use to shield sympathetic populations from removal [16], while her study of the Trump era documents how the same tools were redeployed to withdraw protection from populations previously insulated by them [17]. Collectively, the literature establishes oscillation, and it underscores the centrality of informality, since many priority shifts occur through guidance rather than statute.

Yet this body of work is primarily institutional and conceptual. It persuasively explains why presidential control matters, but it leaves underdeveloped the empirical mapping between executive instruments and measurable shifts in adjudication. The gap is not one of inattention to consequences. Rather, the gap is that the literature has not consistently operationalized volatility as a pattern that can be measured and then explained through identified institutional channels.

D. Adjudication Inside the Department of Justice

Additionally, a second major body of scholarship concerns the institutional placement of immigration adjudication. Stephen Legomsky, a professor of law at Washington University in St. Louis and a former chief counsel of United States Citizenship and Immigration Services, demonstrates in a foundational account that the immigration courts, housed within the Department of Justice and supervised by political appointees, lack the structural independence that adjudicatory neutrality ordinarily presupposes, and he proposes wholesale restructuring on that basis [18]. Adding to this, the American Bar Association’s Commission on Immigration reached the same conclusion after a comprehensive system review, recommending an independent Article I immigration court [19]. Furthering this, empirical work has since put numbers behind the concern. Analyzing an original dataset of more than 830,000 removal proceedings, Catherine Kim and Amy Semet, professors of law at the University of North Carolina and the University at Buffalo, found that immigration judges’ likelihood of ordering removal varied measurably with the sitting administration’s enforcement posture, with the same appointees deciding cases differently under different presidencies [20]. The finding matters here for a precise reason, illustrating that political transitions reach how charged cases are decided, and reach it even when the governing legal standard is constant.

Connected to this, institutional oversight complements the normative and empirical accounts by documenting persistent operational strain. Government Accountability Office reporting on the Executive Office for Immigration Review emphasizes backlog growth alongside workforce, performance, and data management problems [3], [21]. Those findings carry a significance the reports themselves do not draw out. Delay stretches cases across election cycles, which increases the likelihood that a single case will be adjudicated under several policy regimes, so that backlog becomes a conduit through which reversibility produces lived uncertainty. The institutional literature, however, tends to stop at description, documenting crisis conditions without consistently connecting them to the doctrinal and administrative devices that allow leadership transitions to alter adjudication so rapidly.

E. Attorney General Review and Doctrinal Whiplash

Ultimately, the connection between institutional dependence and legal change is clearest in the certification mechanism, through which the Attorney General can refer immigration cases to himself and issue nationwide precedent. The administrative closure episode illustrates the mechanism with unusual clarity. In Matter of Castro-Tum, Attorney General Sessions sharply restricted immigration judges’ authority to administratively close cases, removing a docket tool that had permitted coordination with collateral relief processes [22]. Three years later, in Matter of Cruz-Valdez, Attorney General Garland overruled Castro-Tum and restored the prior understanding, relying in part on adverse appellate decisions [23]. The certification mechanism has long troubled scholars working at the intersection of administrative and immigration law, most notably Jill Family, a professor of law at Widener University, and the closure episode vindicates that concern by exemplifying a doctrinal whiplash in which interpretive stability is coupled to political turnover rather than to durable institutional criteria [24]. The lesson is structural, illustrating that immigration law can change in a system-wide manner without congressional action, through politically appointed supervision inside the Department of Justice. What remains missing is the downstream accounting, because even when scholarship diagnoses whiplash, it rarely connects reversals to indicators such as time-to-decision or the distribution of outcomes across courts.

F. Empirical Evidence of Outcome Volatility

Adding to this, empirical research supplies direct evidence that outcomes swing. The canonical study is Refugee Roulette, in which Jaya Ramji-Nogales, Andrew Schoenholtz, and Philip Schrag, professors of law at Temple and Georgetown, documented enormous disparities in asylum grant rates across immigration courts and across individual judges within the same court, disparities so large that the authors likened the system to a game of chance [25]. Continually, case-level reporting by the Transactional Records Access Clearinghouse, a data research organization then based at Syracuse University, extends the picture longitudinally, documenting substantial variation over short time windows and sharp changes in grant rates that occur without statutory amendment [2], [4], [5]. Similarly, policy research institutions including the Migration Policy Institute located asylum volatility within a system under strain well before the most recent surge, emphasizing the interaction between processing choices at the border and adjudication capacity, and that earlier diagnosis is used here as historical framing rather than as a description of present conditions [26]. Collectively, these findings matter because they establish the magnitude of volatility that any stability-oriented account must explain, whether or not they demonstrate causation by themselves. The empirical work is strong at documenting swings and pressures, yet it is weaker at integrating mechanism evidence from practice, meaning the kind of evidence that explains how attorneys and clients actually experience those swings as governance conducted through guidance and venue assignment.

G. Background Allocation Rules and the Pressure on Discretionary Pathways

Finally, a distinct strand of policy analysis addresses long-term allocation rules, including the statutory ceiling that caps any single country at seven percent of preference visas and generates chronic queues regardless of any single administration’s priorities. Congressional Research Service analysis documents the scale of the resulting backlog, most visible in the employment-based categories, and models how the ceiling shapes it [27]. The relevance here is indirect but important, illustrating that persistent queues in lawful pathways increase the practical salience of discretionary and humanitarian alternatives that sit closest to executive control and are therefore most exposed to reversal. Yet per-country caps are usually discussed as an equity problem in legal migration rather than as a component of enforcement volatility, even though the resulting pressures shape who is channeled into the system’s least stable pathways.

H. The Gap

Read together, the literature establishes four propositions with unusual consistency. First, prosecutorial discretion is foundational and largely insulated from review. Second, Presidents govern immigration through enforcement instruments. Third, adjudication lacks structural independence. Fourth, outcomes fluctuate at magnitudes visible in public data. What the literature has not supplied is an integrated account of how these design choices combine to produce enforcement instability as a predictable feature of the system, and of how that instability moves through adjudication to reach similarly situated individuals across time and place. Doctrinal and institutional work explains why reversal is legally easy and diagnoses the dependence that transmits it, yet it seldom measures the downstream effects. Conversely, empirical work measures those effects, yet it seldom traces them to the institutional channels that produce them.

Two adjacent literatures are deliberately set aside, and the choice deserves a word of explanation. The constitutional debate over plenary power concerns the outer limits of congressional and executive authority over immigration rather than the stability of enforcement within those limits. Additionally, the federalism scholarship on state and local participation in enforcement raises significant questions of its own, but the volatility examined here originates at the federal center. Bracketing these strands keeps attention fixed on the specific interaction that existing work leaves unexplained, and the contribution of this evaluation accordingly builds upon the established scholarship rather than positioning itself against it.

Ultimately, this evaluation works across the seams identified above by integrating doctrinal analysis with longitudinal indicators of volatility and original mechanism evidence from a key informant interview. The quantitative indicators identify temporal discontinuities and cross-adjudicator variance that require explanation, while the qualitative evidence tests whether the operational pathways emphasized in the literature are observed in practice by an actor positioned to watch policy change move through the system. The aim is not to restate that immigration is political. Rather, the aim is to specify how statutory delegation, institutional dependence, and limits on review interact to make reversibility routine, and to use that account as the predicate for the stability-oriented reform analysis in Part V.

III. Methodology

This evaluation adopts an explanatory sequential mixed-methods design, described by John Creswell and Vicki Plano Clark, methodologists writing on research design, as a structure in which quantitative analysis is conducted first and qualitative evidence is then used to account for the quantitative patterns rather than merely to accompany them [28]. The premise is straightforward. Immigration governance produces publicly observable outcomes that fluctuate over time, and those fluctuations can be described with longitudinal indicators, yet they cannot, standing alone, explain the institutional and legal mechanisms that produce them. The design therefore moves in two stages, from documented volatility to mechanism-based explanation, and it does so in service of the research question stated in Part I, namely why enforcement policy shifts sharply when the statute does not, and what that instability does to fairness in adjudication.

A. The Quantitative Phase

To begin, the quantitative phase assembles longitudinal indicators that are plausibly sensitive to administrative change within the immigration court system. The scope is deliberately narrow. Both indicators are drawn from adjudication before the Executive Office for Immigration Review, and neither captures enforcement activity at the Department of Homeland Security or affirmative adjudication at United States Citizenship and Immigration Services, a limitation returned to in Part III.D.

The first set of indicators concerns adjudication capacity and delay. Pending-caseload figures for the Executive Office for Immigration Review, as reported by the Government Accountability Office and as tracked in TRAC’s backlog data, serve a dual analytic function [2], [3], [21]. Backlog operates as an outcome variable and as a measure of system strain, and it operates additionally as a structural mechanism, because when a backlog expands, the temporal distance between charging and adjudication grows, and that distance exposes a single proceeding to more policy regimes over its life. Delay, in other words, is the conduit through which reversals reach people, and it is selected as an indicator for precisely that reason. Importantly, the two sources compile pending caseload differently, since GAO reports figures derived from EOIR’s own reporting while TRAC compiles case-by-case records obtained under the Freedom of Information Act, and TRAC’s totals have historically run higher as a result. The two series are therefore reported separately in Part IV.A rather than joined into a single continuous trend line.

Adding to this, the second set of indicators concerns adjudicatory outcomes, with an emphasis on asylum grant rates and completion volumes derived from TRAC’s case-level reporting [4], [5]. Asylum adjudication is a domain in which priority-setting, rulemaking choices, and adjudicator behavior interact in ways that can move outcomes rapidly without statutory amendment. Grant-rate measures are used here as descriptive indicators of volatility, and they are never treated as self-proving evidence of causation. Their function is narrower and more defensible, namely to identify discontinuities, administration-linked inflection points, and the dispersion of outcomes across adjudicators that together frame the explanatory task for the qualitative phase.

B. The Qualitative Phase

Additionally, the qualitative phase consists of a key informant interview with an experienced immigration attorney practicing in a large metropolitan area of the northeastern United States, referenced throughout as [Redacted]. The interview is designed as mechanism evidence rather than as a survey of practitioner opinion. Its purpose is to test whether the causal pathways emphasized in the literature and suggested by the quantitative patterns are actually observed in practice by an actor positioned to watch policy change move through the system at the case level.

The interview instrument was semi-structured and targeted concrete operational channels through which leadership transitions are experienced, including shifts in enforcement prioritization and prosecutorial discretion practice, changes in asylum adjudication behavior, docket management, and the timing effects produced by administrative delay. The analysis codes the interview thematically for recurring mechanism claims capable of explaining, rather than merely restating, volatility in the quantitative indicators. Ultimately, four themes emerged and are carried forward into Part IV.B: policy change implemented through informal guidance; variation in outcome across venues and adjudicators; the narrowing of practical discretion produced by docket pressure; and the rule-of-law consequences of unpredictability for similarly situated individuals. Other channels raised in the interview, including the transfer of cases between courts, are not analyzed here because the interview did not yield mechanism evidence on them sufficient to support a claim.

C. Ethics and Confidentiality

Several protections structure the qualitative component, and they warrant explicit statement. The interview was conducted under a nondisclosure agreement, and the informant is anonymized throughout as [Redacted]. He gave informed consent to participate and to the use of quoted material, and he is described only in deliberately general terms, limited to professional role, general areas of practice, region of practice, and approximate length of practice. That generality is by design, as the description is intended to convey the informant’s vantage point without permitting identification, and no further identifying detail is disclosed anywhere in this evaluation. Direct quotations were shared with the informant before inclusion and, in places, have been lightly edited for clarity under his review, with those edits confined to concision and syntax rather than to the substance of any statement. Where precision in the informant’s own words was inessential to the mechanism claim, careful paraphrase is used instead. This evaluation was undertaken independently and was not subject to institutional research ethics review. The protections described above, including informed consent, a nondisclosure agreement, anonymization of the informant, and pre-publication review of all quoted material, were adopted to meet the standards ordinarily applied to key informant research.

D. Limitations

Ultimately, four limitations warrant candor. First, a single key informant supplies mechanism evidence, so that the interview can illustrate how volatility is transmitted in practice, yet it cannot establish how widely any particular experience generalizes across the bar. Second, the informant’s practice extends to affirmative matters adjudicated by United States Citizenship and Immigration Services, whereas both quantitative indicators are drawn from immigration court proceedings, meaning that portions of his vantage point sit outside the range of the measured data. Third, the quantitative indicators are descriptive, identifying discontinuities that coincide with administrative transitions yet unable, standing alone, to isolate the effect of a leadership change from confounders such as shifts in the composition and volume of cases arriving at the border. That confounder is a serious one for the fiscal year 2025 backlog figures in particular, and Part V.C treats it directly rather than setting it aside. Fourth, the most recent outcome data, particularly the 2025 grant-rate and backlog figures, remain provisional and subject to revision as additional cases are decided and reported, and interruptions in federal data releases in late 2025 mean that the most recent available case-level records run only through August 2025. Collectively, the design mitigates these limits through triangulation, as doctrine identifies the levers, the indicators show the movement, and the informant connects the two, so that no single strand is asked to carry the argument alone.

IV. Results

This Part reports the findings of both phases in the sequence the design requires, presenting the quantitative indicators first and the practitioner-reported mechanisms second. Interpretation of these findings, including assessment of competing explanations for them, is reserved for Part V.

A. Quantitative Findings

1. Pending caseload

To begin, the pending caseload before the immigration courts grew substantially across the period examined, and the two available sources report that growth on different bases. The Government Accountability Office found that EOIR’s pending caseload stood at roughly 1.8 million cases at the start of fiscal year 2023, more than triple the figure at the start of fiscal year 2017, and it recorded persistent workforce, performance, and data management problems across the same period [3]. Separately, TRAC’s backlog tracking indicates that the caseload surpassed 3.5 million by mid-2024 and peaked near 3.7 million at the close of fiscal year 2024 [2]. As noted in Part III.A, these figures derive from different compilation methods, since GAO reports EOIR’s own counts while TRAC compiles case-by-case records obtained under the Freedom of Information Act, and they are therefore reported here as two series rather than as one. What both series record in common is direction, illustrating sustained and historically unprecedented growth across the period from fiscal year 2017 through fiscal year 2024.

Continually, fiscal year 2025 produced a reversal of that direction. The TRAC-tracked backlog fell to roughly 3.4 million by August 2025, constituting the first sustained decline in more than a decade [2], [5]. Three proximate conditions appear alongside that decline in the reporting: new case filings fell sharply, monthly completions accelerated, and the Department terminated dozens of sitting immigration judges while appointing none of its own during the fiscal year [2], [5]. Additionally, the decline did not correspond to a comparable reduction in delay, since the average pending case had still been waiting well over 600 days [2].

2. Asylum grant rates

Adding to this, asylum grant rates declined across the same period. TRAC data show the grant rate peaking at 52.6 percent in September 2023 and then falling steadily, to 38.2 percent by August 2024, to 35.8 percent that October, and to 19.2 percent by August 2025 [4], [5]. Measured from the September 2023 peak, that represents a decline of roughly sixty-three percent. Two measurement cautions attach to the figures. First, the two TRAC reports drawn upon here measure across different windows and characterize the decline differently in their own headline framing, so the sixty-three percent figure calculated from the peak should not be conflated with the smaller declines those reports describe across their respective shorter periods [4], [5]. Second, TRAC’s own reporting records that the decline proceeded steadily across the change of administrations rather than breaking sharply at the transition [5].

3. Completion pace

Continually, the pace of adjudication did shift abruptly at the transition, in contrast to the rate trend just described. Monthly asylum completions roughly doubled in the spring of 2025, peaking above 12,000 in April and May against the 6,000 to 7,000 that had prevailed before [5]. This indicator therefore moves discontinuously where the grant rate moves continuously.

4. Dispersion across adjudicators

Finally, outcome variation was recorded across adjudicators as well as across time. TRAC’s most recent judge-level reporting records that within the San Francisco immigration court alone, individual judges’ grant rates ranged from 4.8 percent to 97.1 percent [5]. That measure captures dispersion among adjudicators inside a single court rather than variation between courts. It is consistent in kind with the disparities documented in Refugee Roulette nearly two decades earlier [25].

B. Qualitative Findings

The key informant, [Redacted], is an experienced immigration attorney practicing in a large northeastern metropolitan area, with more than two decades of experience representing asylum seekers, family-based petitioners, and applicants for lawful permanent residence. Consistent with the protections described in Part III, he is characterized no more specifically than that, and his quoted statements have been reviewed by him and lightly edited for clarity with his approval. Four themes were identified through thematic coding of the interview, and they are reported here in the order they are taken up in Part V.

To begin, the first theme concerns the speed and informality of policy change. “Policies can change overnight,” the informant reported. He described governance conducted through memoranda and supervisory directives that require neither notice-and-comment procedures nor a transition period, and he reported that such changes force attorneys to advise clients under conditions where expectations can be invalidated without warning. Additionally, he reported that the government’s posture in removal proceedings can harden or soften without any formal change in law.

Adding to this, the second theme concerns variation in outcome by assignment. The informant reported that “two clients with nearly identical facts can receive opposite outcomes depending on where their case lands.” He attributed this variation to differences in local implementation, docket pressure, and adjudicator practice rather than to differences in the governing standard.

Continually, the third theme concerns the practical effect of docket pressure on discretion. The informant reported that when dockets expand, adjudicators have less time per case, and that discretion narrows in practice even as it remains formally broad. He described the observable consequences as fewer continuances, thinner hearings, and greater pressure to resolve cases quickly, and he reported that these conditions fall hardest on respondents who need time to gather evidence, obtain counsel, or pursue collateral relief.

Ultimately, the fourth theme concerns client perception of the system’s legitimacy. In the informant’s words, “for many clients, the law feels secondary to politics.” He connected this perception to the combination of frequent policy change and long pendency, under which a respondent’s hearing arrives whenever it arrives and is governed by whichever standard is then in force. His summation of the stakes was as follows: “the system does not need to be generous to be just, but it does need to be stable.”

C. Summary of Findings

Collectively, the results record movement of substantial magnitude in every measured indicator across a period in which the statutory standard did not change. Pending caseload grew without precedent and then reversed direction within a single fiscal year. Asylum grant rates fell by roughly sixty-three percent from their peak, though continuously rather than discontinuously. Completion pace roughly doubled within months of a transition. Dispersion among adjudicators inside a single court spanned nearly the entire available range. Alongside these figures, the practitioner evidence reports four operational mechanisms, namely informal and immediate policy change, outcome variation by assignment, the narrowing of practical discretion under docket pressure, and a resulting perception among clients that political posture governs rather than law. Part V turns to what these findings mean and to which explanations they will and will not support.

V. Discussion

The results reported above establish that the immigration system’s outputs move at substantial magnitude without statutory change, and that practitioners observe identifiable channels through which such movement reaches individual cases. This Part interprets those findings. It proceeds in five steps, first identifying the delegation that makes reversal inexpensive, then the institutional dependence that transmits reversal into adjudication, then the competing explanations that the quantitative findings will and will not support, then the consequences for fairness, and finally the reforms that follow.

A. Delegation and Low-Friction Instruments

To begin, the Act is dense with substantive grounds of admission, removal, and relief, yet it does not decide how the system will be run day to day. It does not specify how charging decisions will be distributed across categories of noncitizens, how scarce detention capacity will be allocated, or how enforcement attention will be sequenced across time and place. Those tasks are left to the Executive, and they are left in a manner that invites discretionary governance rather than rule-bound administration.

Opposingly, this delegation is often defended as practical necessity, and the defense has force, since enforcement operates across millions of individuals with finite resources and Congress cannot draft a statute that prespecifies every enforcement decision. Necessity, however, is only the first step in the argument. The harder question is how the system channels discretion once granted, because a system can acknowledge prioritization while still insisting on stabilizing commitments, such as durable procedural baselines or meaningful review of certain categories of change. Ultimately, the modern immigration system does relatively little of this disciplining work, and that omission explains why enforcement can be reorganized sharply across administrations without statutory revision.

Connected to this, the doctrine reviewed in Part II supplies the enabling conditions, and its combined effect can be stated briefly. Heckler and Reno place charging and prioritization at the discretionary core and wall them off from ordinary review [6], [7], [8]. Arizona centralizes enforcement judgment in federal hands, entrenching the Executive as the system’s practical manager rather than its mere implementing agent [9]. United States v. Texas demonstrates that priority guidelines addressed to large populations are difficult even to bring before a court [10]. Collectively, instability does not require that executive choices be unlimited, requiring only that the main levers of change be hard to contest, and the doctrine plainly satisfies that condition.

Furthering this, the scholarship on presidential administration explains what follows. When policy is produced through enforcement instruments rather than statute, the legal system supplies fewer built-in frictions against reversal [13], [14]. A statutory compromise binds until Congress repeals it, whereas a guidance-based regime binds until someone signs a new memorandum. In a system organized around executive tools, instability therefore becomes a function of electoral turnover rather than a failure of administration. The procedural discipline described in Part II.B qualifies this picture without changing it, illustrating that Regents and Biden v. Texas govern the manner of reversal while leaving its frequency untouched [11], [12]. The first qualitative finding reported in Part IV.B maps directly onto this architecture, since an informant who reports that policies change overnight is describing the operational face of a system in which the binding instrument is a memorandum rather than a statute.

B. Dependent Adjudication

Additionally, delegated enforcement discretion would not produce volatility of this magnitude if adjudication were insulated from executive control, and it is not. Immigration judges sit within the Department of Justice under supervisory structures that differ sharply from Article III courts and from most independent adjudicatory bodies [18]. The Attorney General retains authority to set binding nationwide precedent through certification, and Department leadership shapes the docket-management tools that govern the pace and structure of proceedings. Collectively, these institutional facts allow enforcement priorities to migrate into adjudication through management and precedent, even when formal legal standards remain unchanged.

Ultimately proving this in full, a real world example demonstrates the channel at work, and the analytic point concerns what the reversal did to litigants rather than the bare fact that it occurred. Castro-Tum did more than change a case-management technique [22]. Administrative closure had allowed judges to pause proceedings while collateral relief unfolded elsewhere, so restricting it forced those cases back onto saturated dockets, altered the bargaining environment in removal proceedings, and shifted leverage toward the government without any change in substantive law. Cruz-Valdez then restored the prior framework three years later [23]. The sequence reveals a distinctive instability, in that a politically appointed official can alter nationwide adjudicatory tools through a precedent decision with minimal procedural friction, and a later appointee can restore them, so that the cycle tracks political transition rather than adjudicatory evolution.

Connected to this, commentators have framed this dependence primarily as a legitimacy problem, and it is one [18], [19], [24]. Crucially, it is also a stability problem, and the empirical record reviewed in Part II bears that out, since the same immigration judges decided cases measurably differently under different sitting administrations [20]. When the adjudicatory forum itself responds to political posture, the system applies different procedural conditions across time in addition to different priorities, because continuance practice, the pace of hearings, and the availability of pauses for collateral relief all shape outcomes by shaping the opportunity to present claims. Additionally, the dependence changes incentives inside the system, as a judge whose docket tools can be redefined by supervisory decision operates on contingent ground rules, and a practitioner advising clients must treat the procedural terrain as unstable. Ultimately, those are predictable consequences of housing adjudication inside a department that answers to electoral change.

C. Interpreting the Indicators, and What They Will Not Support

Furthering this, the quantitative findings reported in Part IV.A support some explanations and not others, and the distinction requires statement rather than suppression.

To begin with the backlog reversal, no statutory amendment explains it, yet the three conditions recorded alongside it in Part IV.A.1 do not point uniformly toward administrative recalibration. Terminating adjudicators reduces capacity, which should depress completions rather than accelerate them, so the acceleration is most plausibly attributable to changes in how cases were resolved rather than to how many adjudicators remained. More significantly, the sharp decline in new filings reflects conditions at the border as much as decisions inside the immigration courts, and border flows are exactly the confounder identified in Part III.D. Ultimately, the honest reading is therefore narrower than a claim of pure administrative causation. What the reversal demonstrates is that the system’s most basic output measure changed direction within months of a transition, through instruments that required no statutory revision, and that the composition of that change was itself set by executive choices about filing, staffing, and disposition.

Opposingly, the grant-rate findings resist an even simpler reading. Because the decline proceeded continuously across the change of administrations rather than breaking at it, the figures will not support a step-function account tied to a change in party [5]. That is a genuine constraint on the argument advanced here, and it is reported as such. What the figures do establish is magnitude, illustrating that outcomes moved at a scale demanding explanation in any account that takes equal treatment seriously, and that if the statute did not change, then something else did.

Connected to this, the completion-pace finding reported in Part IV.A.3 is the strongest single administrative signature in the dataset, precisely because it moves discontinuously where the grant rate does not. A doubling of monthly completions within months of a transition is difficult to attribute to caseload composition and straightforward to attribute to managerial direction, since throughput is exactly the variable that docket management controls. That distinction between rate and pace refines the argument rather than weakening it, locating the clearest evidence of executive influence in how quickly cases are pushed through rather than in how they come out. Additionally, it converges with the third qualitative finding, in which the informant reported that docket pressure narrows practical discretion, since accelerated throughput is the mechanism by which that narrowing occurs.

Finally, the dispersion finding reported in Part IV.A.4 speaks to a different dimension of the same problem. A range from 4.8 percent to 97.1 percent among adjudicators inside one court illustrates that the practical meaning of eligibility attaches to the adjudicator who receives the case, and it converges precisely with the second qualitative finding, in which the informant reported that clients with nearly identical facts receive opposite outcomes depending on where their case lands. Collectively, the temporal and adjudicator-level findings shift the meaning of legal eligibility itself, in that eligibility remains a threshold requirement while the practical chance of success attaches to the procedural regime in force at the moment of adjudication and to the adjudicator assigned. Ultimately, the indicators therefore do more than supply descriptive context, illustrating that instability is large enough to constitute a defining feature of governance rather than a residual imperfection.

D. Consequences for Fairness and the Rule of Law

Collectively, a stable legal system does not require identical outcomes across all cases, since immigration adjudication necessarily includes discretion and humanitarian relief will always involve judgment. The rule-of-law demand established in Part I is more basic, requiring that people be able to predict the procedural conditions under which their cases will be adjudicated, and that similarly situated individuals not be separated by timing, assignment, or shifting administrative posture. On Fuller’s account, that is a demand of congruence and constancy rather than of generosity, and it is one a restrictionist system can satisfy [29], [30].

Ultimately, the structural features traced above undermine those conditions in combination. Broad delegation allows policy to be made through instruments that can be replaced quickly, while adjudicatory dependence allows the resulting managerial and precedential shifts to reshape procedure and doctrine from within an executive department, subject to judicial correction only at the procedural margins. The combined effect is that fairness becomes contingent, illustrating that for a respondent, success can depend on whether the case is heard before or after a reversal, on which adjudicator receives it, and on whether a docket tool such as administrative closure happens to exist at that moment. The fourth qualitative finding is the predictable residue of that arrangement, since a client who concludes that the law feels secondary to politics is drawing a rational inference from the conditions the first three findings describe. Motomura’s account illustrates why the inference is not irrational, since for a large population the operative question has long been executive tolerance rather than formal status [15], and the redeployment of protective tools that Wadhia documents illustrates how quickly that tolerance can be withdrawn [16], [17].

Additionally, the instability carries distributional consequences, and they fall in one direction. Unpredictability burdens those with the fewest resources most heavily, because a well-resourced respondent can adapt, whether by retaining counsel quickly or by pursuing several relief pathways at once against the chance that one survives a reversal, whereas a respondent with limited resources experiences delay as compounding hardship and rapid change as a direct threat. When a legal system routinely places the cost of instability on those least able to bear it, that instability becomes a fairness problem in its own right.

E. Implications for Reform

Ultimately, the four reforms below follow from the findings rather than from abstract preference, and each aims to raise the cost of reversal where reversibility is doing the most harm while preserving lawful executive judgment where prioritization is unavoidable. Each additionally answers a specific mechanism identified in Part IV.

First, Congress should codify limited procedural baselines that currently rest on withdrawable guidance. The first qualitative finding illustrates that instability enters most destructively through docket tools and informal implementation devices that can be revoked without transitional safeguards. A statute might therefore codify access to administrative closure in defined circumstances, and might in the same measure clarify the standards for continuances tied to collateral relief and impose minimum transparency requirements on large-scale prosecutorial discretion guidance. None of this would eliminate discretion. Rather, it would move key procedural conditions from informal governance onto a statutory footing, and that shift matters because it changes the legal default, so that under a codified baseline an administration may still redirect enforcement resources without as easily unsettling the ground rules that determine whether respondents can meaningfully present claims.

Second, structural reform of immigration adjudication should be evaluated as a stability intervention in addition to a legitimacy intervention. Converting EOIR into an independent Article I tribunal is usually defended in terms of the professionalism and perceived fairness that independence is thought to secure [18], [19]. The findings here add a related point, illustrating that independence also functions as a stabilizer. So long as adjudication sits inside the Department of Justice, a change in political leadership can quickly reshape docket management and precedent through certification [22], [23], and the completion-pace finding reported in Part IV.A.3 shows how immediately that reshaping registers in throughput. An Article I structure would not erase politics from immigration, yet it would change how easily politics can reach the adjudicatory tools that shape outcomes, and that change matters most in a system with long timelines where cases routinely span administrations. Opposingly, the honest counterargument is cost, since transition would be disruptive and independence carries accountability tradeoffs of its own. Ultimately, those are one-time and manageable costs weighed against a recurring structural harm.

Third, the system needs a more disciplined approach to variation across adjudicators and courts. Both the dispersion finding and the second qualitative finding identify assignment as a major determinant of outcome [5], [25]. A stability-focused response need not require national uniformity in every discretionary judgment. Rather, what it requires is greater consistency in the administration of procedural standards and in the public articulation of enforcement criteria. The Departments of Justice and Homeland Security can advance this by publishing transparent performance metrics, by tying training to those published criteria, and by releasing court-level and judge-level data at intervals that allow deviations to be detected. Collectively, the goal is narrower than identical results across courts, aiming instead to shrink the degree to which two similarly situated individuals are separated by local implementation differences that cannot be justified in legal terms.

Fourth, Congress should address the background allocation rules that magnify volatility by channeling applicants into discretionary pathways. Per-country limits and related queue structures generate chronic delay for whole categories of applicants, which increases reliance on forms of relief that sit closest to executive recalibration [27]. Reforming those rules would not directly solve enforcement instability, yet it would relieve the pressure that pushes individuals into the system’s least stable channels. This is a feasibility point as much as a fairness point, illustrating that even modest statutory adjustments which reduce extreme backlogs shrink the class of cases pending long enough to experience several policy regimes.

Ultimately, these proposals share a common analytic logic. The findings do not support the claim that discretion can be eliminated, and none of the reforms attempts it. Rather, they support a narrower claim, namely that a legal system can choose where discretion operates and how much friction governs its reversal. Procedural baselines, adjudicatory independence, transparency, and queue relief each work by increasing continuity in the conditions that matter most for fair adjudication, and each respects the institutional reality that enforcement will always involve prioritization.

VI. Conclusion

This evaluation asked why immigration enforcement policy shifts so sharply across administrations when the governing statute does not change, and what that instability does to the fairness of adjudication. The answer developed across the preceding parts is that instability is a foreseeable consequence of legal and institutional design rather than an accidental byproduct of political disagreement. Congress has delegated expansive enforcement authority through the Immigration and Nationality Act, adjudication remains structurally dependent on politically appointed leadership within the Department of Justice, and judicial doctrine has treated core prioritization decisions as largely insulated from review while the review that does exist disciplines the manner of reversal rather than its frequency.

The results support that account with appropriate qualification. Pending caseload, completion pace, and dispersion among adjudicators all moved at magnitudes that no statutory amendment explains, and the completion-pace finding in particular moved discontinuously at a transition in a manner difficult to attribute to caseload composition. Opposingly, the grant-rate decline proceeded continuously rather than breaking at the transition, and the backlog reversal is entangled with border flows that this design cannot isolate, so neither finding will bear a simple partisan reading. Alongside the figures, the practitioner evidence identifies the channels through which movement reaches individual cases, describing a system in which policies change overnight, in which clients with nearly identical facts receive opposite outcomes depending on assignment, and in which docket pressure narrows discretion in practice while it remains broad on paper.

Collectively, the contribution is to connect two literatures that have largely run in parallel, the doctrinal and institutional work explaining why reversal is legally inexpensive, and the empirical work measuring the movement that reversal produces. Immigration law will remain politically contested, and research has clearly outlined that contestation need not entail procedural whiplash, because a system can change its priorities without unsettling the ground rules under which pending cases are heard. Ultimately, the reforms proposed above do not promise agreement about who should be admitted or removed. Rather, they promise something more modest and, for the person whose case spans three administrations, something more important, which is that the rules of the road will still be there when the case is finally called.

VII. References

1. [1] Immigration and Nationality Act, 8 U.S.C. §§ 1101–1537.
2. [2] Transactional Records Access Clearinghouse (TRAC), Immigration Court Backlog Tool, https://tracreports.org (last visited Jan. 12, 2026).
3. [3] U.S. Gov’t Accountability Off., GAO-23-105431, Immigration Courts: Actions Needed to Address Workforce, Performance, and Data Management Challenges (Apr. 26, 2023).
4. [4] Transactional Records Access Clearinghouse (TRAC), Asylum Grant Rates Decline by a Third (Nov. 2024), https://tracreports.org/reports/751/ (last visited Jan. 12, 2026).
5. [5] Transactional Records Access Clearinghouse (TRAC), Immigration Court Asylum Grant Rates Cut in Half (Nov. 2025), https://tracreports.org/reports/766/ (last visited Jan. 12, 2026).
6. [6] Heckler v. Chaney, 470 U.S. 821 (1985).
7. [7] Reno v. American-Arab Anti-Discrimination Committee, 525 U.S. 471 (1999).
8. [8] 8 U.S.C. § 1252(g).
9. [9] Arizona v. United States, 567 U.S. 387 (2012).
10. [10] United States v. Texas, 599 U.S. 670 (2023).
11. [11] Department of Homeland Security v. Regents of the University of California, 591 U.S. 1 (2020).
12. [12] Biden v. Texas, 597 U.S. 785 (2022).
13. [13] Adam B. Cox & Cristina M. Rodríguez, The President and Immigration Law, 119 Yale L.J. 458 (2009).
14. [14] Adam B. Cox & Cristina M. Rodríguez, The President and Immigration Law Redux, 125 Yale L.J. 104 (2015).
15. [15] Hiroshi Motomura, Immigration Outside the Law (2014).
16. [16] Shoba Sivaprasad Wadhia, Beyond Deportation: The Role of Prosecutorial Discretion in Immigration Cases (2015).
17. [17] Shoba Sivaprasad Wadhia, Banned: Immigration Enforcement in the Time of Trump (2019).
18. [18] Stephen H. Legomsky, Restructuring Immigration Adjudication, 59 Duke L.J. 1635 (2010).
19. [19] American Bar Association Commission on Immigration, 2019 Update Report: Reforming the Immigration System (2019).
20. [20] Catherine Y. Kim & Amy Semet, An Empirical Study of Political Control over Immigration Adjudication, 108 Geo. L.J. 579 (2020).
21. [21] U.S. Gov’t Accountability Off., GAO-17-438, Immigration Courts: Actions Needed to Reduce Case Backlog and Address Long-Standing Management and Operational Challenges (June 1, 2017).
22. [22] Matter of Castro-Tum, 27 I. & N. Dec. 271 (A.G. 2018).
23. [23] Matter of Cruz-Valdez, 28 I. & N. Dec. 326 (A.G. 2021).
24. [24] Jill E. Family, Administrative Law Through the Lens of Immigration Law, 64 Admin. L. Rev. 565 (2012).
25. [25] Jaya Ramji-Nogales, Andrew I. Schoenholtz & Philip G. Schrag, Refugee Roulette: Disparities in Asylum Adjudication, 60 Stan. L. Rev. 295 (2007).
26. [26] Migration Policy Institute, The U.S. Asylum System in Crisis: Charting a Way Forward (2018).
27. [27] Congressional Research Service, R46291, The Employment-Based Immigration Backlog (2020).
28. [28] John W. Creswell & Vicki L. Plano Clark, Designing and Conducting Mixed Methods Research (3d ed. 2018).
29. [29] Lon L. Fuller, The Morality of Law (rev. ed. 1969).
30. [30] Joseph Raz, The Rule of Law and Its Virtue, 93 Law Q. Rev. 195 (1977).